Professional Documents
Culture Documents
Fracture-Calibration Tests
M.J. Mayerhofer,* Mining U. Leoben; C.A. Ehlig-Economides, SPE, Schlumberger;
and M.J. Economides, SPE, Texas A&M U.
Summary In tight formations with permeabilities < 1 md for oil and 0.01 md
Fracture-calibration pressure decline has been used for determina- for gas, a conventional well test cannot be performed because the
tion of the leakoff coefficient, a bulk variable describing the process reservoir may not even flow without hydraulic fracture stimulation.
of fluid influx into the reservoir, normal to the created fracture face. However, reservoir permeability is an essential parameter for deter-
In this work, the fluid loss is modeled in terms of the controlling mining the optimum fracture size. 9 Mayerhofer et al. 10 - 12 published
mechanisms: flow through the filter cake, the invaded zone, and the results of a series of experiments to show the nature of flow through
reservoir. A rigorous model describes unsteady-state fluid flow the fil ter cake.
Our paper describes a model that decouples the reservoir and filter-
from fractures of varying area into the formation, with the filter cake
cake behavior as flow in the formation from an infinite-conductivity
considered as a time- and rate-dependent skin effect. The injection
fracture 13 with a rate- and time-dependent skin effect analogous to the
history is superposed on the pressure decline. This work provides a
Cinco-Ley and Samaniego-V.14 fracture-face-damage skin effect.
straight-line technique for determination of reservoir permeability
In modem well-test interpretation, log-log diagnostic plots of the
and fracture-face resistance. Log-log diagnostic plots provide the
pressure change and the derivative function are used to distinguish
means to recognize visually whether the transient response is domi-
various flow regimes, which develop characteristic patterns. Once
nated by flow in the reservoir or at the fracture face. We found that
the flow regime is identified, specialized plots are used to compute
the pressure transient very frequently is dominated by the flow in the
important well or reservoir parameters. The governing filtration
reservoir rather than through the filter cake. The reservoir perme-
model can be analyzed in the same manner. By estimating the vari-
ability (an essential value for fracture design that is usually not
able leakoff rate for fracturing-fluid leakoff during fracture closing,
available) can be estimated, while the model captures all trends of
the rate-normalized pressure and its derivative can be used as the
the falloff-pressure variation.
diagnostic tool for flow-regime identification.
+ [~(qFj
L A . _ A. ~]
qFj-I) ytn - t j _1 Ro
Gas Reservoir
I!.Ptace I!.PR I!.p
j=m p.} p.}-I (ftlmd) (psi) (psi) (psi)
At k=0.01 md 0.3 284 921 1,205
[ ~(
3 2,840 921 3,761
- L qlj -
~
qlj-I ) j tm
A-:- - tj _ 1] At k=0.1 md 0.3 284 290 574
j= 1 p.} p,}-I 3 2,840 290 3,130
At k=1 md 0.3 284 91 375
3 2,840 91 2,931
+ (2ap,ufJrRo/Ap,m)(RD.nqFn - RD,mqlm)' ..... (10)
Initial reservoir pressure, Pi, cancels out. Eq. 10 can be solved itera- face resistance (Ro = 0.3 and 3 filmd, respectively). The calculations in
tively for the pressure change, I!.p(tn), by use of qFj (Eq. 9) for any Table I were made for the same totalleakoff volume during pumping.
time tn > tm, thereby providing the pressure falloff vs. time. Inserting Results for the oil reservoir show that for permeability up to I md,
Eq. 9 and the leakoff rates during pumping (Eq. 7) into Eq. 10 leads the pressure gradient in the reservoir dominates. Only an inordinately
to an expression that can be solved iteratively for I!.p(tn) and tn. high fracture-face resistance (laboratory values for crosslinked poly-
Eq. 10 satisfies the physics of filtration and linear elastic fracture mer fluids were found to range between 0.3 and 1.5 filmd after 20
mechanics while preserving the material balance. It is related to the minutes of filtration JO) and a permeability > I md can result in a
physical properties Ro and k, but, unlike the G-function, it does not dominant skin component. In the case of a gas reservoir, the skin com-
require knowledge of a constant leakoff coefficient at any time. ponent may dominate for permeabilities as low as 0.01 md, depend-
Traditionally, specific leakoff coefficients have been postulated for ing on the value of the resistance. And, yet, as a case study described
separate phenomena, such as those controlled by compressibility and later and results of other field applications of the technique show, the
viscosity. In those cases, individual zones have been combined as reservoir and not the filtercake resistance often dominates, even in gas
conductances in series. A simple (harmonic average) and some more- wells. This suggests that field filtercake resistances are much lower
complicated averaging techniques l5 have been used to calculate the than laboratory-derived values.
combined leakoff coefficient. Instead, Eq. 10 addresses the individual This analysis shows that the individual components can be quan-
pressure gradients (Eq. I) in their correct relative contribution, and tified and related. As Table 1 shows, the relative contributions can
the components are added as resistances in series. This approach is be distributed at any ratio, dependent on the reservoir (often primar-
straightforward because the solutions are given by well-known mod- ily) and the filtercake properties. This result is not apparent when the
els that have been used frequently in well-test applications. concept of totalleakoff coefficient is used and is not in accord with
other publications. 8,15
Relative Contributions of the Pressure Components Fracture-face resistances in Table 1 are laboratory-derived (filter-
Eq. 6 is used to calculate the total pressure gradient from the fracture cake resistance) and thus should be considered as ideal, maximum
(assumed to be equal to that in the weUbore) into the reservoir and the values that assume even distribution and deposition of the filter cake
pressure gradients of the individual components during pumping. The throughout the created-fracture walls. In reality, such a deposition is
first term of the right side of Eq. 6 represents the pressure drop in the unlikely, resulting in an even greater reservoir influence on leakoff.
formation. In the last term, the filter cake and the filtrate-invaded An important implication from the results in Table 1 is the indica-
zone are combined as a fracture-face skin effect or resistance (I!.Pjace tion that this analysis may determine whether fluid-loss additives
in Eq. I). Resistance Ro is the reference resistance at the end of pump- are required in the design of the main fracture treatment. For homo-
ing. Table 1 presents pressures calculated after 20 minutes of pump- geneous, lower-permeability reservoirs, artificialleakoff control is
ing. Reservoir permeability varied from 0.1 to 10 md for an oil and an unnecessary expense. This consideration would allow lowering
from 0.01 to I md for a gas reservoir, with a low and a high fracture- of the fracturing-fluid viscosity only enough to achieve the required
---........
'in
~
Interpretation Method
c..
c:
k=O.1 md
~
:
. . .........
.. '
~
The following outlines the sequence of interpretation .
1. Data acquisition. Acquire pressure-falloff data, preferably in
.0
~ ______ ~ 11._,:... small sample intervals.
~""'1'
to)
c:
.a unit slope 2. Computation and plotting. Compute and plot the log-log diag-
III
>
_,l:'; I k= 10 md nostic plot (with a reasonable estimate for Af based on injected vol-
ume) of ll.Pm and its derivative function vs. time to assess the rela-
+=i
cv 100 ............ .. ........1............ .......... ................ ..
> tive importance of the reservoir and skin pressure-drop behavior
c
III and to detect any deviations from expected behavior as seen on the
"C
"C
sensitivity plots in Figs. 1 through 3. (These could be the result of a
c: change in the fracture area during closing or of perforation-friction
cv - - - pressure
c.. pressure drop unaccounted for during the injection phase.)
<I
... - derivative 3. Simulation. Simulate the pressure-derivative function with the
estimated area and reasonable assumptions of Ro and permeability,
and correct the initial estimate of fracture area by matching the ear-
10 ly-time derivative data.
4. Specialized plot. Estimate reservoir permeability and fracture-
1 10
face resistance.
time (min)
5. Simulation. Match the pressure change and derivative function
in a log-log plot and compare the simulated pressure history with the
Fig. 1-Pressure and derivative plot: influence of permeability. measured one.
- . - . - k= 1 md iAPrn
-- --
.,:.~.:~:'1~::
'iii 'iii
a.. 0.
Q) 0.1 Q)
> > 0.1
~ ~
>
.;:: >
.;::
i
"0 "0
E
0.01
~c:
. , ,. ,'r 0.01
c: ; I' .. ' ,
tU ,. ,.... ""'-L tU
E !derivative
derivative
.% ." .%
0.001 0.001
0.1 1 10 100 0.1 10 100
time (min) time (min)
Fig. 2-Rate-normalized plot: influence of reservoir permeability. Fig. 3-Rate-normalized plot: influence of fracture area.
Q)
> -----------------------(------------------------r--------------------------- ....
(I)
0.1 ::I
~
>
'':::
Q)
. ~
I/)
I/)
....
(I)
0-
1000 -------7 ! closure
G-function straight-line
"0
ic:
til
0.01 .__. j.............
1
800
600
o 0.5 1 1.5 2 2.5
G-function
0.001
Fig. 8-Gas reservoir case study: comparison of the G-function
0.1 1 10 100 straight line and response calculated by the model in this paper.
time (min)
Recent work by Gu et al.!6 demonstrates the interest in perme-
Fig. 5-Rate-normalized pressure plot: gas reservoir case study. ability determination. However, Gu et al.'s suggestion that impulse-
test analysis can be applied to injection tests in the same manner as
it applies for much shorter injection times requires consideration of
the length of time needed to establish pseudoradial flow. When the
fracture created by the injection test is ~ 50 ft, it may take several
Ap= 82,000 ftl
hours, or even days, to observe the onset of pseudoradial flow. Yet,
~1000
a.
'-' Ro= 0.56 ftlmd the linear flow response may occur (as in the second field case study
c in this work) within minutes of closure. In that case, analysis after
0 k= 0.003 md closure offers independent confirmation of kx/.
'zj
u
C
.....::I Conclusions
(I)
> This paper demonstrates that analysis of the pressure decline fol-
'zj
IV 100 lowing the injection test in a fracture-calibration treatment can yield
>
''::: the formation permeability, the filtercake resistance, the area of the
(I)
"0
created fracture, and the reservoir pressure.
"0 The analysis uses the log-log diagnostic plot of the rate-normal-
C
IV ized pressure calculated with leakoff rates computed from material
0- A p= balance and fracture compliance considerations and the llPrn deriv-
<l
k= 0.002 md ative with respect to the logarithm of elapsed time between the end
of pumping and fracture closure. Sensitivity studies have shown that
10 this plot clearly indicates when the treatment behavior is dominated
1 10 100 by the effects of the filter cake and when it is dominated by the tran-
time (min) sient-pressure response associated with linear flow in the formation.
The reservoir permeability and the fracture-face resistance are com-
Fig. 6-Gas reservoir case study: match of pressure change and puted by use of a specialized plot with axes defined to provide a
derivative straight line for all the data acquired during the pressure decline up
to the time of fracture closure. The results are verified on a plot of
~
7000 the pressure change and its derivative that shows that the measured
'iii
~
a. data match a simulation of the pressure response by the model for
pressure decline provided in this paper.
~
c 6000 ---
m= 1460 Values for permeability and fracture-face resistance frequently
ci
ct:
E are not, and often cannot be, determined by any other procedure.
.
rT
c
ci
5000 -b= 2678
Nomenclature
ct:
u.
c
Af= fracture area (one face), L2 ft 2
0-
:::::- 4000 Ap= permeable fracture area, L2. ft2
a. RO= 0.56 ftlmd
b= intercept in Eq. 7
<l
q= fracture compliance, U 2/m, psi-!
: :